The client approached Oodles to build an AI-powered video processing system capable of detecting safety infractions on live RTSP camera streams. The objective was to overlay real-time PPE compliance indicators on video feeds while supporting large-scale concurrent processing. The solution was designed to operate on local Linux infrastructure or AWS, generate new RTSP streams with AI overlays, and provide an administrative portal for managing users, cameras, reports, and system preferences.

The scope of work included designing a scalable architecture to process multiple RTSP camera streams and apply AI-based PPE detection in real time. The client required generation of new RTSP streams with visual overlays or caption-only output based on latency conditions. Oodles was responsible for implementing latency monitoring and fallback logic, containerized deployment, and modular system design for future API integrations. The project also included development of an admin portal for user management, camera assignment, live previews, reporting, and server configuration.

Oodles delivered a modular AI video processing solution that ingests RTSP streams, applies PPE detection using the client-provided AI model, and rebroadcasts enhanced streams with overlays or captions. The system supports concurrent camera processing, latency-based mode switching, and deployment via Docker for local or cloud environments. An admin web portal was developed to manage users, camera streams, reports, and storage preferences. The solution was structured to support future scaling, additional detection models, and integration with external control systems.